Transcript SpeechTek West 2007-Improvement in the quality of

Improvement in the quality
of automated dictation by
making explicit use of
semantic knowledge
Klaus Stanglmayr
Friday, February 23, 2007
Philips Speech Recognition
Systems
Klaus Stanglmayr, Friday, July 17, 2015
2
Philips Speech Recognition Systems
• A business unit of Royal Philips
Electronics
• 12 years continuous speech
recognition
• The world‘s most successful speech
recognition technology for
professionals
- 8000 sites
- 50 nations
• Large network of integration partners
(200+)
• On the forefront of technological
innovation
Klaus Stanglmayr, Friday, July 17, 2015
3
Bringing sense and simplicity to life
Klaus Stanglmayr, Friday, July 17, 2015
4
Speech is the most
common and
easiest means of
communication
between people...
...and technology!
Klaus Stanglmayr, Friday, July 17, 2015
5
Speech Recognition in healthcare
Klaus Stanglmayr, Friday, July 17, 2015
6
Healthcare is the world’s largest service sector
13%
Worldwide Healthcare Expenditure:
US$3,300 bln in 2002
2%
North America
8%
Western Europe
49%
Japan
China/Hong Kong
North America makes up 49%
of the worldwide spent
Rest of the World
Source: Medistat
28%
Healthcare Expenditures will grow from ~8% of worldwide GDP to ~10% in 2010
10%
8%
6%
4%
2%
0%
Source: WHO
1950
2000
Klaus Stanglmayr, Friday, July 17, 2015
2010
74
Medical technology continues to transform Healthcare
Imaging
earlier diagnosis saves lives and reduces costs
Minimally invasive surgery
reducing patient trauma and costs
Healthcare IT
Right Information at the right time
enables better treatment and lower costs
Molecular Medicine
Preventing disease from happening
Around 70% of the survival improvement in heart attack mortality is a
result of changes in technology.”
Cutler & McClellan, 2001
Sources: Russ Coile, Futurescan 2003, SG-2
Klaus Stanglmayr, Friday, July 17, 2015
8
Klaus Stanglmayr, Friday, July 17, 2015
9
Hands-free Interaction in the Hospital
• Advanced voice control
technology
• Enables clinicians to interact
with medical equipment when
their hands are already
occupied
Klaus Stanglmayr, Friday, July 17, 2015
10
Clinical Documentation – primarily by speech
• Primary use of Speech
Recognition in healthcare is
for clinical documentation
• A large chunk of patients
health records was dictated;
free text
• Clinician dictates as usual
• Speech recognition
technology creates a draft
that is reviewed edited by
medical transcriptionist using
keyboard or speech
• Transcript is returned to
clinician for review and
authentication and insertion
into record
Klaus Stanglmayr, Friday, July 17, 2015
11
Medical Transcription
Technology replaces Medical Transcription Costs
20
15
Billion $ 10
Transcription
Costs
5
Technology
0
2005
Future
• Estimated $18 to $25 billion industry
• Estimated 300,000 to 400,000 MTs in US
Klaus Stanglmayr, Friday, July 17, 2015
12
Documentation is
the lifeblood of healthcare
Klaus Stanglmayr, Friday, July 17, 2015
13
Complex process of caring for patients
Clinical documentation comprises the essential
building blocks for all healthcare processes
Each step must be accurately and thoroughly
documented
Delivering care is a complex process with many
potential failure points.
Errors in medical care are usually not due to
negligence by medical staff
Physicians, when surveyed, indicate that they are
most interested in capabilities
- that reduce documentation time
- positively affect their communications with
patients, other physicians, and payers
- provide clinical information to minimize their
time away from non-patient functions
Klaus Stanglmayr, Friday, July 17, 2015
14
Challenges in Healthcare
Diminished quality of
healthcare
-Inadequate information
-Illegible entries
-Misinterpretations
-Lack of shareable
information
-Lack of uniform capture
Klaus Stanglmayr, Friday, July 17, 2015
15
NLP/NLU
• One of the tasks that an NLP system can perform is the parsing of a
sentence to determine its syntax. Determining the semantic
meaning of a sentence is a much harder task to perform. When an
NLP system can achieve this, we can talk about Natural Language
Understanding (NLU).
• "Understanding" language means, among other things, knowing
what concepts a word or phrase stands for, and knowing how to
link those concepts together in a meaningful way. It's ironic that
natural language, the symbol system that is easiest for humans to
learn and use, is hardest for a computer to master. Long after
machines have proven capable of inverting large matrices with
speed and grace, they still fail to master the basics of our spoken
and written languages.
Klaus Stanglmayr, Friday, July 17, 2015
16
SPEECH UNDERSTANDING
• Understands the entire context of dictation and recognizes the
physician’s intent, while adapting to a variety of speaking styles
and dictation habits
• Technology has the flexibility to handle the way people really
speak—with varying accents, dialects, styles, and speeds
• Converts free text into structured information
• Organizing Meaning
• Technology gleans meaning from spoken words and organizes
it automatically into meaningful categories
• Automatically structured documents enable efficient review
• Automatically adds normals, templates, macros
• Encoding Data
• Services process dictation, encode clinical facts, like
medications, dosages, allergies, measurements like blood
pressure readings, and much more.
• Semantic Interoperability
Klaus Stanglmayr, Friday, July 17, 2015
17
Intelligent Speech Interpretation (ISI)
Challenges for traditional speech recognition systems
• human speech/human speakers are not perfect
•
•
•
•
•
•
•
punctuation omitted
disfluent speech
hesitations
pauses
silences
redundant speech
repetitions/corrections
Focus on optimizing speech recognition: “make
it work in the “real” world”
Klaus Stanglmayr, Friday, July 17, 2015
18
Challenges – A perfect speaker?
POSTOPERATIVE_DIAGNOSIS
[AAHHMMMMM] Status post complete oral and dental rehabilitation PERIOD
NEXT_PARAGRAPH [OOHHMMMMM]
[NEXT SECTION]
OPERATION PERFORMED Complete oral and dental rehabilitation PERIOD
NEXT_PARAGRAPH
[cough]
ANESTHESIA
General anesthesia by the anesthesia staff PERIOD NEXT_LINE
Duration of surgery: Close to 45 minutes.
Incision: none.
NEXT_PARAGRAPH
[NEXT SECTION]
[________________][paper rustling] [________________]
FINDINGS [AAHHMMMMM] The patient is a 22-year-old male with missing teeth and permanent dentition
COMMA mental retardation COMMA seizure disorder COMMA microcephaly COMMA and
gastroesophageal reflux disease PERIOD Since [__] March first 2004 COMMA there has not been much
change in the appearance of the patient PERIOD [AAHHMMMMM] The patient also had periodontal
disease evidenced by moderate bone loss COMMA hyperplastic gingiva and to three to seven millimeter
pockets PERIOD
NEXT_PARAGRAPH
This is the end of dictation, thank you
Klaus Stanglmayr, Friday, July 17, 2015
19
Understanding more...
Intelligent Speech Interpretation
Dictated Text
Recognized Text
Standard phrases
End of dictation. Thank you.
Does not appear in final document.
Redundant phrases
Send copy of report to
Does not appear in final document.
Section headings
Condition on/at/upon/of/ discharge
Next is condition...
Next section is condition..
Condition on Discharge
Dates
May five two thousand two
May fifth two thousand two...
May 5, 2002
Automatic
punctuation
No chills fevers night sweats weight loss...
No chills, fevers, night sweats, weight loss...
Silences/pauses
There has not been (------pause----) much change...
There has not been much change.
Non-speech dictation
There has not been (paper rustling) much change...
There has not been much change...
Hesitations
There has not been (AAHHMMMMM) much...
There has not been much...
Contraction
There hasn‘t been
There has not been
Orthographic variants
She is here today because she has a letter from her (obgyn)
resident from the university...
She is here today because she has a letter from her
Ob-Gyn or ob/gyn or ob-gyn or Ob/gyn
(customizable) …
Klaus Stanglmayr, Friday, July 17, 2015
20
Final Report
POSTOPERATIVE DIAGNOSIS
Status post complete oral and dental rehabilitation.
OPERATION PERFORMED
Complete oral and dental rehabilitation.
ANESTHESIA
General anesthesia by the anesthesia staff.
Duration of surgery: Close to 45 minutes.
Incision: none.
FINDINGS
The patient is a 22-year-old male with missing teeth and permanent dentition, mental retardation, seizure
disorder, microcephaly, and gastroesophageal reflux disease. Since March 1st 2004, there has not been
much change in the appearance of the patient. The patient also had periodontal disease evidenced by
moderate bone loss, hyperplastic gingiva and to 3-7 millimeter pockets.
John Smith, MD
Philips Hospital Vienna
Triester Straße 64
A1101 Vienna, Austria
Klaus Stanglmayr, Friday, July 17, 2015
21
All truth passes through three stages.
First, it is ridiculed. Second, it is
violently opposed. Third, it is accepted
as being self-evident.
Arthur Schopenhauer
For further information please
contact:
[email protected]
Tel.: +43-1-60101-4110
http//www.philips.com/speechrecognition
Klaus Stanglmayr, Friday, July 17, 2015
22